Can't unstack the result of multidimension interp

[zxdawn]

Hi, I'm trying to interpolate a DataArray by two coordinates. However, each input coord is two dimension.

So, I stack the input coords first to make them become one dimension and then unstack it back.

import xarray as xr
import numpy as np

da = xr.DataArray(
        np.sin(0.3 * np.arange(40).reshape(5, 4, 2)),
        [("x", np.arange(5)), ("y", [0.1, 0.2, 0.3, 0.4]), ("var", [1, 2])],
)

x = xr.DataArray([[0, 0.5], [1.5, 3]], dims=['x', 'y']).stack(z=['x','y'])
y = xr.DataArray([[0.15, 0.25], [0.3, 0.35]], dims=['x', 'y']).stack(z=['x','y'])

da.interp(x=x, y=y, method='nearest').unstack('z')

Error:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [2], in <module>
     10 y = xr.DataArray([[0.15, 0.25], [0.3, 0.35]], dims=['x', 'y']).stack(z=['x','y'])
     13 # da.sel(x=x, y=y, method='nearest')
---> 14 da.interp(x=x, y=y, method='nearest').unstack('z')

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/dataarray.py:2196, in DataArray.unstack(self, dim, fill_value, sparse)
   2136 def unstack(
   2137     self,
   2138     dim: Union[Hashable, Sequence[Hashable], None] = None,
   2139     fill_value: Any = dtypes.NA,
   2140     sparse: bool = False,
   2141 ) -> "DataArray":
   2142     """
   2143     Unstack existing dimensions corresponding to MultiIndexes into
   2144     multiple new dimensions.
   (...)
   2194     DataArray.stack
   2195     """
-> 2196     ds = self._to_temp_dataset().unstack(dim, fill_value, sparse)
   2197     return self._from_temp_dataset(ds)

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/dataset.py:4211, in Dataset.unstack(self, dim, fill_value, sparse)
   4209         result = result._unstack_full_reindex(dim, fill_value, sparse)
   4210     else:
-> 4211         result = result._unstack_once(dim, fill_value, sparse)
   4212 return result

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/dataset.py:4065, in Dataset._unstack_once(self, dim, fill_value, sparse)
   4062     else:
   4063         fill_value_ = fill_value
-> 4065     variables[name] = var._unstack_once(
   4066         index=index, dim=dim, fill_value=fill_value_, sparse=sparse
   4067     )
   4068 else:
   4069     variables[name] = var

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/variable.py:1695, in Variable._unstack_once(self, index, dim, fill_value, sparse)
   1690     # Indexer is a list of lists of locations. Each list is the locations
   1691     # on the new dimension. This is robust to the data being sparse; in that
   1692     # case the destinations will be NaN / zero.
   1693     data[(..., *indexer)] = reordered
-> 1695 return self._replace(dims=new_dims, data=data)

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/variable.py:985, in Variable._replace(self, dims, data, attrs, encoding)
    983 if encoding is _default:
    984     encoding = copy.copy(self._encoding)
--> 985 return type(self)(dims, data, attrs, encoding, fastpath=True)

File ~/miniconda3/lib/python3.9/site-packages/xarray/core/variable.py:2630, in IndexVariable.__init__(self, dims, data, attrs, encoding, fastpath)
   2628 super().__init__(dims, data, attrs, encoding, fastpath)
   2629 if self.ndim != 1:
-> 2630     raise ValueError(f"{type(self).__name__} objects must be 1-dimensional")
   2632 # Unlike in Variable, always eagerly load values into memory
   2633 if not isinstance(self._data, PandasIndexingAdapter):

ValueError: IndexVariable objects must be 1-dimensional